Passive Input Current Harmonics Filter - Design & Calcuations

[jordon]

Application: SCR-controlled Battery Charger.

Input power: 440 VAC, 60 Hz, 3 ph, 3-wire, 10 KW.
Output power: 42.3 VDC, 200 Amps.
Input Transformer: 440V pri. Δ, 44V sec. Y, 18-pulse (9-phase).

Links shows sketch of application and transformer winding configuration.
Link

I thought the 18-pulse transformer would eliminate most of the unwanted input current harmonics...
However, I'm still seeing harmonics, 5th, 11th, 17th, and 19th are all above 3% (5th and 17th being the highest, at 6%).

I would like suggestions on how to attenuate these harmonics to below 3% using a passive (RLC) AC filter on the 440V primary side.

Design papers, calculations, or rules of thumb are welcome.
Looking for a solution that is robust (allowable ~5-10% tolerance on selected RLC components).

Thanks.

 

[dpc]

These harmonic currents you are "seeing" - are these actual measurements or by calculation?

 

[ozmosis]

Why not look at an active solution? If you know the amperage required to cancel the specific orders mentioned, you only need to select one rated for this current rating whereas a passive solution (in series) needs to be rated for the full 200 amps.

 

[jordon]

These harmonic currents you are "seeing" - are these actual measurements or by calculation?

They are measured values using a power analyzer, phase-locked to the input voltage supply.

Why not look at an active solution? If you know the amperage required to cancel the specific orders mentioned, you only need to select one rated for this current rating whereas a passive solution (in series) needs to be rated for the full 200 amps.

Active solution cannot be considered, sorry.

 

[ScottyUK]

An 18-pulse rectifier design of only 10kW output is an unusual piece of equipment: normally multi-phase rectifiers aren't seen until the high hundreds of kW or into the megawatt range. Even a 6-pulse 10kW rectifier isn't likely to cause much upset to a utility company, unless there are dozens of these things in parallel or the source has an abnormally high impedance. What's unusual about this installation?

FWIW, 10kW is within the reach of a big switchmode supply, which would give you the option of an active front end with a near-perfect input waveform.

 

[sibeen]

I'd be looking at your thyristor firing methodology. With a 18 pulse drive you shouldn't be seeing any appreciable lower order current harmonics.

The other possibility is the incoming voltage. Is it a nice clean waveform and balance well across the three phases?

 

[dpc]

You might have a look at this:

http://www.mtecorp.com/pages/wp-content/uploads/18pulse.pdf

Probably not totally unbiased, but indicates some possible limitations to 18 pulse drives. As ScottyUK stated, it is unusual to see 18-pulse configuration on a drive this small.